KR20230070917A - Electric motor failure online diagnosis system by Machine learning based magneto register current sensor - Google Patents

Abstract

The present invention relates to a method for detecting motor faults online in real time by an artificial intelligence device which has learned the pattern of the current flowing in a defective motor. While operating a motor used in a power plant or industrial site, various defects are created inside the motor, and the fault is reproduced. A current sensor is installed in a motor power supply circuit, the sensor and an artificial intelligence device for learning are connected, and faults are experienced through information received from the current sensor and stored as data, so that the learned artificial intelligence device diagnoses the fault of the motor online in real time based on learning.

Description

Electric motor failure online diagnosis system by Machine learning based magneto register current sensor}

The present invention relates to a method for detecting motor failures online in real time by an artificial intelligence device that has learned the pattern of current flowing in a defective motor. It relates to a method for diagnosing motor failures online in real time through experience and learning of data accumulated by acquiring and storing field reproduction through a current sensor installed in a motor power line.

Motor failures are varied and complex. The fault current is large enough for the protection relay to operate, so while there are obvious faults, there are also very minute faults that may have occurred. In particular, this minute failure does not have a significant effect on motor operation, so there are cases where it does not have a special effect on motor operation even after several months or years. However, such a minute and gradually progressing failure gradually weakens the insulation of the motor, eventually destroying the motor and progressing to an accident, causing a lot of economic and social losses. If it is possible to detect a failure that would have occurred in a motor such as partial discharge or interlayer short circuit, it is possible to prevent a failure before it develops into a major accident and significantly reduce restoration costs. do.

In order to detect the above micro-faults, the present invention uses a magneto-resistor-based current sensor. The magneto-resistor is installed in the magnetic field of the current flowing in the busbar and based on the characteristic that the electrical resistance changes as the magnetic field changes, the magnetic disk and It has been used for reading using the magneto resistance phenomenon in the same data storage technology, and has been applied to various magnetic sensors because its magnetic field resolution is superior to Hall devices.

As described above, the conventional magneto resistor technology relates to a method for manufacturing a magneto resistance element with improved giant magneto impedance characteristics disclosed in Korean Patent Publication No. 10-2011-0115774 and a magneto resistance element manufactured according to the method, and a simple process of irradiating ferromagnetic material with ions. Since it is manufactured using, magnetic sensitivity and magneto impedance characteristics are improved, it is applied so that it can be usefully used for magnetic heads of high-sensitivity information recording media and magnetic sensors for detecting minute magnetic fields. A technology that mainly applies a technology in which electromagnetic characteristics are changed by a magnetic field acting from the outside by applying an impedance sensor has been developed.

However, in order for a magneto resistor-based current sensor to measure current and estimate it as a valid value, it must have various signal processing processes. In particular, noise processing is important for online real-time diagnosis of rotating bodies such as motors, and noise can be measured at maximum. It increases as the current value increases, but in the case of the conventional Hall sensor-based current sensor, the noise value is about 1 to 3% of the maximum measurable current value, so it is possible to track and measure devices that consume a lot of current and devices that use little current at the same time. It becomes very difficult to analyze or measure very minute current changes in devices that use a lot of current, and as a result, monitoring and prediction of devices using machine learning (artificial intelligence learning) have become impossible.

Republic of Korea Patent Publication No. 10-2011-0115774 (published on October 24, 2011) Republic of Korea Patent Registration No. 10-1492825 (registration date February 6, 2015)

The present invention has been made to solve the above problems, and the failure of the motor is determined by integrating the unique pattern of the current flowing through the motor for each failure, the rise of a specific frequency, phase characteristics, harmonic characteristics, waveform distortion, etc. Based on the learning of the motor, it is classified and recognized based on learning, and if a once-recognized failure is experienced again during motor operation, it can be judged as a failure as already learned. By continuously learning the normal state of the motor while operating the motor, If a state is detected, it is recognized as a defect and notified. If a learned failure situation occurs, it is notified as a failure as naming during the learning process. In addition, it was invented for the purpose of preventing destruction and accidents of electric motors by using this.

Meanwhile, other unspecified objects of the present invention will be additionally considered within the scope that can be easily inferred from the following detailed description and effects thereof.

In order to achieve the above object, according to an embodiment of the present invention, the motor abnormality diagnosis based on artificial intelligence learning is performed by installing a magneto-resistor-based current sensor in a power line supplying power to the motor, and measuring the current sensor measured from the current sensor. By acquiring, storing, and learning various current information (signals), an artificial intelligence device that recognizes the information pattern of the load current flowing when a motor failure occurs can diagnose the presence or absence of a motor abnormality based on the learned information.

In the present invention, when a micro-failure such as partial discharge or interlayer short circuit occurs in the motor, the fault can be prevented before it develops into a major accident by detecting the fault online in real time, and the recovery cost can be significantly reduced through such detection. there is

Figure 1. Measurement diagram of a magneto-resistive-based current transducer (current sensor)
Figure 2. Sensitivity waveform of sensor to be applied to pattern recognition of the present invention
Figure 3. Integrated signal processing block diagram of the present invention
Figure 4. Overall configuration of the system used in the present invention
Figure 5. Signal-to-noise (SNR) ratio of the present invention
Figure 6. Functional block diagram of the sensor module used in the present invention
7. Conceptual diagram of a current sensor module to be applied to the present invention
Figure 8. Conceptual diagram of deep learning data processing (Stacking) of the present invention
Figure 9. Mathematical representation of Training of the present invention

Hereinafter, the present invention will be described in detail.

In order for a device equipped with artificial intelligence to diagnose motor abnormalities based on learning, it starts with measuring current using a magneto-resistor-based current sensor. The current sensor has a cylindrical ring structure with a height of less than 100mm, the size of a finger, which can be easily installed on the wire by opening one side of the current sensor when installed on the power supply line.

The principle of current measurement is that the magnetic field formed around the bus line through which the current flows has a characteristic that the strength of the magnetic field changes according to the magnitude of the current. Current is measured based on this changing principle. The above current sensor is installed on the bus that supplies power to the motor, and the resistance value of the magneto-resistive element inside the current sensor changes in proportion to the change in the magnetic field formed by the current flowing in the bus. A constant voltage is applied, and the correlation between the resistance size and the primary current is formulated and calculated as shown in FIG. 9 based on the applied voltage and current information flowing through the device, and mathematically inversely calculates the current flowing in the primary bus. Measure the current value of the secondary bus bar. The appearance and installation of the current sensor are shown in FIG.

In addition, in order to diagnose motor abnormalities based on artificial intelligence learning, current information measured from a magneto-resistor-based current sensor installed on the power line that supplies power to the motor is acquired and stored, and the load current that flows when a fault occurs in the motor by using the stored data. After comparing and analyzing the pattern of the machine, it is recognized, and the artificial intelligence device diagnoses motor abnormalities based on the learned information.

Micro-failures of motors can be classified and recognized based on learning about each fault by integrating the unique pattern of the current flowing through the motor for each fault, the rise of a specific frequency, phase characteristics, harmonic characteristics, and waveform distortion. If a device equipped with artificial intelligence experiences a once-recognized failure again while operating the motor, it judges it to be a failure as it has already learned. In addition, since the normal state of the motor is continuously learned during motor operation, if an abnormal state other than normal is detected, it is recognized as a defect and notified through a smart device. informs the result of

The motor abnormality diagnosis device by machine learning learning is a ring shape with a size of about tens of mm, that is, a quarter of the size of the palm, and a cylindrical current sensor and a current sensor with a height of a finger that has a structure in which one side can be opened. It is largely composed of a signal processing unit (microcontroller) that receives a current signal from and processes the signal.

In order for a current sensor to measure current and estimate it as a valid value, various signal processing processes are required, and processing for noise is important. Noise increases as the maximum measurable current value increases, but conventionally used Hall sensor-based current sensors have a noise value of about 1 to 3% of the maximum measurable current value, so devices that consume a lot of current and devices that use little current It becomes very difficult to simultaneously track or measure very small current changes in current-consuming devices, and as a result, it becomes impossible to monitor and predict devices using machine learning.

As an example, although the load current of the motor reaches 300A, there is also a very small current pattern of several mA in amplitude of the discharge current signal generated during partial discharge, so existing current converters have a current of several tens of A and a current of several mA. It is difficult to detect changes. This fact is why the sensitivity of a current sensor is so important for load learning.

2 shows two types of current sensors that recognize a current pattern of mA at a high current load of 300 A, and the sensitivity characteristics of the magneto resistor-based current sensor to be applied in the present invention are shown in (2) of FIG. When a load is in operation, the sensitivity of the current sensor, which can detect minute changes in the size of 1/1000 of the load, is a successful factor in subdividing and recognizing the current information flowing in the load. The magneto-resistive used in the system of the present invention As shown in (2) of FIG. 2, the based current sensor shows characteristics suitable for recognizing fine patterns of current changes.

The pattern recognition technology of the present invention applies a technology for diagnosing failures by learning and storing patterns of current waveforms for each motor defect, naming, and storing them in a neural network circuit. One of the aggregation blocks of deep learning is artificial neural networks (ANNs), which originate from information processing in biological neurons. The "deep" part of deep learning is for stacking layers of ANNs to process data. A basic conceptual diagram is shown in FIG. 8 .

In addition, in order to predict and diagnose the state of the motor, it is necessary to go through a training process. The deep learning model shows a mathematical function to learn how to classify data in FIG. The goal of machine learning, or training, is to identify a set of parameters for this function. The parameter search algorithm that Training uses (e.g. Adam, Adagrad, Stochastic Gradient Reduction), the parameter gradient identification algorithm, Back propagation ), STPD (Spike Timing Dependent Plasticity), etc., the above algorithms work together to identify the best parameters and train patterns according to motor failures.

The present invention recognizes a pattern using a time frequency representation (TFR) method that effectively simultaneously expresses the obtained current information in space-time and frequency space. TFR has a complicated calculation method, so it is essential to develop an optimization method to effectively embed it into a microcontroller, and it provides advantages of edge computing such as low-latency, low-power, fast response, resilience to network disturbance, etc. only when it is effectively embedded in a microcontroller. can do. In the case of TFR, the amount of information provided is the largest compared to the amount of given data, so it is possible to increase the accuracy of machine learning while minimizing the amount of computation required and the data storage space to be used in the cloud. The smaller the noise of the current signal, the more precise TFR pattern is obtained. In the case of a signal with less noise when performing the task of recognizing the original current pattern using the most widely used neural network-based reverse image search algorithm for the two TFR patterns It shows a recognition probability of more than 99%, and in the case of a noisy TFR pattern, the recognition probability drops to 80%. It is confirmed that the performance of the TFR expression technique can be optimized when used with the magneto-resistor-based current sensor applied in the present invention. did

In addition, by using an algorithm that recognizes a current pattern that occurs when a device starts operating, it is possible to accurately calculate the starting point of sequential convolution for each device. Based on this unique technology, a technique that accurately calculates the starting point of sequential convolutional operation for each failure is applied to solve the problem of starting sequential convolutional operation at various points in time due to synchronization uncertainty, thereby significantly reducing the amount of computation required for parallel operation. Defects can be identified and recognized

The signal processing unit installed within 10m from the sensor by receiving the current measured by the current sensor through wired/wireless communication uses an analog/digital conversion circuit that converts the analog signal of the current sensor into a digital value and a frequency analysis of the digitally converted current information. It consists of a pattern recognition (signal processing) circuit that recognizes fine patterns, a load recognition neural network circuit equipped with machine learning technology, and a communication circuit that transmits processing information to external smart devices.

The overall configuration of the system showing a block diagram of the sensing-signal processing module to be used in the present invention is shown in FIG. It is responsible for supporting various data communication protocols and has a built-in neural network processing function that is responsible for NILM (non-intrusive load monitoring). It has a built-in function to automatically transmit data to a database on the cloud using various built-in communication support functions, or to visualize and image with various smart devices and inform the results.

Each module consists of a magneto resistor-based current sensor module, an analog input receiving an analog input signal received from the sensor, a signal processing module including a microcontroller composed of a power circuit, amplification and noise removal filtering circuit, and data composed of Bluetooth, Wi-Fi, and Ethernet. It consists of a communication circuit and a remote monitoring system, and the overall configuration of the system is shown in FIG. 4.

The current measurement value of the magneto-resistor-based current sensor is transmitted in analog form to the microcontroller, which is the signal processing unit, and the microcontroller including the A/D conversion unit that converts the analog signal received from the current sensor into digital uses a matched filter to reduce noise. and implements ultra-precision resolution through this noise reduction technology. In addition, in order to minimize resolution deterioration due to noise generated in the analog amplifier embedded in the signal processing unit including the microcontroller, the signal up-conversion function is used in the analog amplifier stage. As shown in FIG. 5, the signal generated by the sensor In the analog amplifier stage, the signal-to-noise ratio (SNR) is greatly improved by raising the frequency band to a high frequency band where flicker noise does not exist, realizing high resolution without being affected by noise.

In order to solve the noise problem, the signal processing unit amplifies the signal at the analog amplifier stage and at the same time applies noise shaping technology to shape most of the noise in the frequency space so that it is included in a specific frequency band. Afterwards, the noise is minimized through an optimization operation (optimal detection) that removes the noise shaped in the digital signal processing circuit. In addition, in the case of offset, it uses a technology to measure and predict the offset value in real time, accurately predicts it, and then compensates for the offset using a feedback circuit in real time. Since the offset value of the sensor continuously changes according to the external environment such as temperature and the aging of the motor itself, the real-time compensation technology developed and applied as described above is applied in the present invention. A functional block diagram of the sensor module used in the present invention is shown in FIG. 6 and a conceptual diagram of the current sensor module is shown in FIG.

The signal processing unit, which includes a microcontroller installed within a certain distance from the current sensor, 10m in an embodiment, not only current data signal processing function, but also machine learning function by neural network, wired, It has a built-in communication chip capable of transmitting signals in the form of wireless, so that signals can be transmitted to external smart devices or monitoring devices.

The processing method of the input electrical signal to be used in the neural network responsible for AI and non-intrusive load monitoring (NILM) using machine learning is a method using data patterns in space-time, but when only space-time data is used, data The amount of useful information that can be extracted compared to the amount of is limited, so a method of using signals in the frequency space together is applied. By applying the time frequency representation (TFR) method, which effectively and simultaneously expresses signals in time and frequency space, edge computing such as reduction in delay, realization of fast response speed, low power consumption, and effective response in emergencies such as communication network down advantages can be provided.

In the case of TFR, the amount of information provided is the largest compared to the amount of given data, so it is a very important technology to increase the accuracy of machine learning while minimizing the amount of computation required and the data storage space to be used in the cloud. A much more precise TFR pattern is obtained when the signal is less noisy. When performing the task of recognizing the original current pattern using the convolutional neural network (CNN)-based reverse image search algorithm, which is the most widely used convolutional neural network (CNN), two TFR patterns can secure a recognition probability of more than 99% in the case of a signal with low noise. , in the case of noisy TFR patterns, the recognition probability fell to 80%. Performance is optimized when the developed TFR expression method is used with a current sensor.

An external smart device notifies the driver of motor failure information based on the signal transmitted from the signal processing unit on site.

Claims (7)

In the motor abnormality diagnosis system based on artificial intelligence learning,
Magneto-resistor-based current sensor installed in the power supply line;
An A/D conversion unit for converting the analog signal measured by the current sensor into a digital one;
A pattern recognition circuit that recognizes fine patterns through frequency analysis of the converted digital signal, a filtering circuit that amplifies signals and removes noise, a neural network circuit equipped with machine learning technology, a microcontroller and communication that processes the signals of the circuit a signal processing unit including a circuit;
Online fault diagnosis system by magneto-resistor current sensor-based motor fault learning, characterized in that consisting of; smart device for displaying the processed signal.
According to claim 1,
The filtering circuit for amplifying the signal and removing noise is characterized in that offset compensation using noise shaping and feedback is possible.
According to claim 1,
The communication circuit is an online fault diagnosis system by magneto-resistor current sensor-based motor fault learning, characterized in that consisting of Bluetooth, Wi-Fi, and Ethernet,
According to claim 1,
A neural network circuit equipped with machine learning technology applies a time frequency representation (TFR) method. An online fault diagnosis system by magneto-resistor current sensor-based motor fault learning.
According to claim 1,
The pattern recognition circuit is characterized by diagnosing partial discharge and interlayer short circuit of the motor including the rise of a specific frequency, phase characteristics, harmonic characteristics, and waveform distortion. Online fault diagnosis system by magneto-resistor current sensor-based motor fault learning.
According to claim 1 or 5,
The pattern recognition circuit recognizes the current pattern occurring at the start of the operation of the motor and calculates the start point for each failure.
According to claim 1,
The current sensor is a magneto-resistor current sensor characterized by applying a voltage of DC 5 to 10V to image a pattern of current based on the voltage applied so that the resistance value of the internal magneto-resistive element is changed and current information flowing through the element. Online failure diagnosis system by based motor failure learning.

Images